Integrated circuit technology can implement various calculation functions in hardware. Addition operation is a core function for implementing the calculation functions. Subtraction, multiplication, and division operations can all be implemented based on the addition operation. Conventionally, the addition operation is binary operation implemented by transistor circuits. A simple one-bit full adder generally needs at least more than ten transistors. Multi-bit addition operations may need more transistors.
Resistive-switching Random Access Memory (RRAM) can store data by resistance variations, and has a wide application prospect as a next-generation non-volatile memory. Typically the RRAM comprises a top metal electrode, a bottom metal electrode, and a dielectric film therebetween having a resistive-switching characteristic. The resistive-switching layer typically comprises metal oxides, such as TiO2, HfO2, ZrO2, Ta2O5, NiO, and ZnO, etc. The RRAM typically has two operation modes, i.e., a unipolar mode and a bipolar mode. In the unipolar mode, the RRAM is switched between a high resistance state and a low resistance state with a voltage that exhibits a same polarity at both ends of the RRAM being applied thereto. In the bipolar mode, the RRAM is switched between the high resistance state and the low resistance state with a voltage that exhibits different polarities at both ends of the RRAM being applied thereto. The RRAM can be used as a data memory, in which each resistance value corresponds to a data value. For example, a high resistance value corresponds to a data value ‘0’, a resistance value lower than the high resistance value corresponds to a data value ‘1’, and a resistance value that is even lower corresponds to a data value ‘2’, and so on.
In a conventional computer, calculation and storage are independent from each other and are implemented by different modules or devices, respectively. If multiary calculations can be implemented by a simple structure and the calculation and the storage can be integrated, more data can be processed than by the conventional addition operation. Accordingly, operation efficiency can be improved.